Our aim is to fill a surprising and glaring gap in the software available today for biomedical simulation: the lack of a standard library for stochastic simulation usable from a variety of programming languages and able to capitalize on various hardware acceleration options. We will produce a free, efficient, portable library for scalable stochastic simulation, featuring selectable algorithms ranging from a fast implementation of the exact algorithm introduced by Gillespie to recent faster but approximate algorithms. The library will feature support for C, C++, Java, Python, Perl, MATLAB and Mathematica under Windows, MacOS, Linux and FreeBSD. Keeping the same high-level front-end API (application programming interface), we will provide multiple back ends suitable for different hardware scenarios: typical desktop single-processor systems, multicore and multiprocessor systems, and FPGA-based (field programmable gate array) hardware acceleration boards. We will also provide an SBML (Systems Biology Markup Language) interface layer, allowing easy, direct simulation of models expressed in SBML. All software, as well as hardware designs and configuration software, will be released under the open-source GNU LGPL license for research use. By introducing such a portable, robust and flexible library, we hope to simultaneously reduce the effort wasted on repeated reimplementation of the same software by different groups, and provide a baseline reference that more advanced researchers can use as a jumping-off point for new and improved algorithm and software development.